3 ms·
This is surprising to me. I would guess that close to 50% of my friends and acquaintances are in the top 5%, and the overwhelming majority are 70-75% percentile
by bkjelden 11y ago
This is surprising to me. I would guess that close to 50% of my friends and acquaintances are in the top 5%, and the overwhelming majority are 70-75% percentile or higher.
I guess that just goes to show you how skewed from the averages the set of people you know and interact with can be.
- cm2012 11y agoOn this note, some stats for perspective: 1) 30% of Americans have a college degree of any kind (inc. associates, etc.) 2) Median household income is $50k yearly nationwide. It's $50k in NYC and $70k in SF. 3) Median single person income is $32k nationwide.
- stygiansonic 11y agoI guess this should not be too surprising. People of the same socioeconomic group tend to associate with each other, because of work, similar interests, etc. This is basically a clustering effect. In terms of living distance from parents, if you live in an area where people tend to flock to (because that's where the jobs are, or other reasons), then it's further likely your friends will be in the same boat as you. More reasons why anecdotes don't necessarily match the overall population statistics. * Note: Massive generalizations in my above statements.
- jczhang 11y agoAnd of course it's not surprising among people who post on Hacker News that they would be among the top 5%. It's just that having friends only in your socioeconomic level will put you in a bubble regardless of how much reading you do about the poverty of others. The level of empathy for others can really only exist if you have personal connections.
- bkjelden 11y agoYea, when I think about it some more, it makes sense. But it is interesting how much of a bubble you can live in without really realizing it. Although the 25 miles in the Great Plains states shown in the article is still surprising to me. As anyone who's lived in those states will attest, 25 miles is nothing.
- Scoundreller 11y agoThe NYT appropriately used median instead of mean to discount the effects of the outliers. I'm sure the average would be a lot higher (because of people moving between coasts).